Research
NFL quarterback passing in freezing cold
Our scoring study found that cold weather barely moves NFL totals — a deliberate null. This is the study underneath that null: every outdoor start since 2006, matched to kickoff temperature at the stadium's own coordinates, and measured against the same quarterback's own average for the same season rather than against the league.
The headline
Across 4,555 outdoor starts by 107 quarterbacks from 2006 to 2025, passing output rises by about 0.30 yards per degree of temperature — a real but shallow gradient that only bites hard at the extreme. Below 20°F a starter throws -21.7 yards against his own season form, with completion rate -2.94 points lower and -0.28 fewer touchdowns — all three clear significance. Ordinary winter cold (32–45°F) costs roughly 8 yards. Interceptions and sacks do not move in any band.
Per degree of temperature
+0.30 yds
+0.07 yards per attempt per 10°F
Fitted across all 4,555 starts · average kickoff 58.2°F
Severe cold (under 20°F)
-21.7 yds
-2.94 completion points · -0.28 TDs
64 starts · t = -2.60
Freezing (20–32°F)
-9.5 yds
-2.16 completion points
214 starts · yardage t = -1.88 sits just under significance
Every temperature band against the quarterback's own season
| Temperature at kickoff | Starts | Passing yards | Yards per attempt | Completion % | Touchdowns | Interceptions | Calm wind only, yards |
|---|---|---|---|---|---|---|---|
| Severe cold (under 20°F)average 14.3°F · wind 9.1 mph | 64 | 213.5-21.7t = -2.60 | 6.72-0.48t = -2.01 | -2.94t = -2.44 | -0.28t = -2.21 | +0.11t = 1.11 | -9.5t = -1.1043 starts |
| Freezing (20–32°F)average 27.5°F · wind 8.6 mph | 214 | 229.8-9.5t = -1.88 | 7.25-0.07t = -0.50 | -2.16t = -3.32 | +0.08t = 0.88 | +0.10t = 1.44 | -5.2t = -0.78139 starts |
| Cold (32–45°F)average 38.9°F · wind 8.4 mph | 749 | 228.5-8.2t = -3.13 | 7.11-0.20t = -2.85 | -1.10t = -3.05 | -0.07t = -1.73 | +0.02t = 0.47 | -3.6t = -1.16523 starts |
| Mild (45–60°F)average 52.8°F · wind 8.1 mph | 1,322 | 237.4+0.2t = 0.10 | 7.26+0.01t = 0.10 | -0.22t = -0.81 | +0.01t = 0.25 | -0.05t = -2.08 | +3.2t = 1.37942 starts |
| Warm (60–75°F)average 67.2°F · wind 7.6 mph | 1,375 | 240.1+3.5t = 1.81 | 7.30+0.05t = 1.01 | +0.66t = 2.58 | -0.01t = -0.47 | +0.03t = 1.16 | +5.6t = 2.591,046 starts |
| Hot (75°F+)average 80.8°F · wind 7.1 mph | 831 | 242.9+5.4t = 2.16 | 7.39+0.14t = 2.10 | +1.03t = 3.13 | +0.08t = 1.97 | -0.02t = -0.49 | +6.2t = 2.17661 starts |
The top figure in each cell is the raw band average; the smaller figure is the gap against the quarterback's own season baseline, with its t statistic. Anything under roughly 2 is noise and is greyed out. The 40-start minimum was fixed before any band was inspected and is applied to every study we publish.
Two honest caveats. First, the warm and hot rows are positive by construction — a quarterback's baseline includes his own cold-weather starts, so the warm and cold rows are mirrors of each other, not independent results. Second, the coldest games are also the windiest (the severe band averages 9.1 mph against 7.1 mph in heat), and the calm-wind column shows how much of the "cold" effect is really wind: in severe cold the yardage gap shrinks from -21.7 to -9.5 once windy games are removed. The remainder points the same way but is no longer significant on its own — we report it as direction, not as a number to price.
Why the per-quarterback baseline matters
League passing has drifted upwards across the window: outdoor starters averaged 209.2 yards at 6.88 per attempt in 2006 and 214.6 yards at 7.11 per attempt in 2025. A study that pooled two decades against a single league average would hand weather the credit for rule changes and offensive evolution. Because each baseline is built inside a single season for a single quarterback, that drift cannot reach the temperature comparison at all.
| Season | Qualifying outdoor starts | Passing yards | Yards per attempt |
|---|---|---|---|
| 2006 | 323 | 209.2 | 6.88 |
| 2007 | 321 | 214.4 | 6.87 |
| 2008 | 350 | 214.9 | 6.81 |
| 2009 | 334 | 219.5 | 6.92 |
| 2010 | 366 | 227.6 | 7.20 |
| 2011 | 365 | 239.8 | 7.26 |
| 2012 | 368 | 233.3 | 7.11 |
| 2013 | 366 | 240.2 | 7.16 |
| 2014 | 384 | 241.2 | 7.26 |
| 2015 | 384 | 252.9 | 7.39 |
| 2016 | 320 | 243.2 | 7.12 |
| 2017 | 319 | 232.0 | 6.96 |
| 2018 | 320 | 244.5 | 7.25 |
| 2019 | 320 | 237.9 | 7.12 |
| 2020 | 313 | 238.4 | 7.14 |
| 2021 | 359 | 232.0 | 7.03 |
| 2022 | 349 | 226.3 | 7.13 |
| 2023 | 358 | 224.8 | 7.02 |
| 2024 | 354 | 223.4 | 7.16 |
| 2025 | 358 | 214.6 | 7.11 |
What holds up
- Severe cold is real. Under 20°F, yards, completion rate and touchdowns all fall significantly against the same quarterback's season form — the only band where every measure moves together.
- Ordinary winter cold is modest. From 20°F to 45°F the yardage cost sits in single digits, and completion rate does more of the moving than volume.
- Mistakes do not rise. Interceptions and sacks sit flat in every band, matching the turnover study and the wind study.
- Wind is the confound to watch. The coldest games are the windiest, and the calm-wind split shows a large share of the severe band's gap travelling with the wind rather than the thermometer.
- This is why scoring stays flat. Cold suppresses passing efficiency; teams compensate elsewhere. The quarterback pays for cold weather more than the scoreboard does.
Method and limits
- Sample: 4,555 outdoor starts, 2006–2025. A start qualifies when the passer threw the most attempts for his team, attempted at least 10 passes, and logged at least 8 such games that season so his baseline is stable. Domes and closed roofs are excluded entirely rather than used as a control.
- Weather: hourly reanalysis at the stadium's own coordinates, temperature taken at kickoff, game-window wind reported alongside every band and used for the calm-wind split (under 10 mph).
- Baseline: leave-one-out mean of the same quarterback's other qualifying starts in the same season, computed separately for yards, yards per attempt, completion rate, touchdowns, interceptions and sacks.
- Not controlled: opponent pass defence, game script and score differential, precipitation, play-calling, and receiver availability. A team that abandons the pass in a freeze shows up here as fewer yards, which is part of the effect we are measuring rather than a separate one.
- No prop or betting lines are attached to these starts, so this study reports performance against a quarterback's own form and makes no claim about whether the market prices cold correctly.
Data generated 2026-09-27. Numbers refresh as the archive tops up; corrections are dated and kept visible.
Related research
- NFL Temperature & Scoring — the deliberate null this study explains: cold moves the quarterback, not the scoreboard.
- NFL QB Passing in Wind — the companion study: wind is the condition that genuinely moves passing, and the confound inside every cold-weather game.
- NFL Turnovers in Bad Weather — why cold and wet football does not mean extra giveaways.
References & data sources
- Open-Meteo ERA5 historical reanalysis — archive-api.open-meteo.com (hourly game-window weather at each venue’s own coordinates)
- ESPN NFL scores and odds — espn.com/nfl (schedules, final scores and consensus closing totals)
- Internal NFL historical game archive — venue identity, roof status and final scores.
- Internal NFL quarterback boxscore archive — per-start passing lines parsed from public boxscores.
- Every dated change we have made to a published study is listed on our corrections log.
Where this is applied
These findings feed the game-time read on every NFL venue page, using the same sample-size and significance rules published here.
OverUnderWeather Research Desk
Independent data analysis on public sports and weather records. Studies match historical game outcomes to archived game-window weather, apply pre-set sample and significance rules, and publish null results when that is what the data show. Editorial policy and corrections log.

Methodology reviewed by Kerry G. KauffmanB.S. in Statistics, Lehigh University
“After examining the complete methodology and structure of the study, I am in agreement that the Welch’s t-test is the appropriate method when comparing two means.”— Kerry G. Kauffman, 2026-09-26
Cite this research
Free to republish with a visible credit and a link to this page. Ready-made citations below — click to copy.
Frequently asked
Does cold weather actually hurt quarterback passing?
Yes, but less than the mythology suggests, and mostly at the extremes. Below 20°F a starter throws about 22 fewer yards than his own season average, with completion rate down roughly 3 points and about a quarter of a touchdown less. From 20°F up through ordinary winter cold the effect shrinks to single-digit yardage — real, but modest.
Why compare a quarterback to himself?
Because comparing cold games to league average would mostly measure who happened to be playing. Every figure here is one start against that same quarterback's own average across the rest of that same season, with the game in question removed from his baseline. That strips out skill, team quality and the league-wide rise in passing volume over twenty years in one step.
Is it the cold, or the wind that comes with it?
Partly the wind. The coldest games are also the windiest, so we report a calm-wind split alongside every band. In severe cold the yardage gap roughly halves when windy games are removed, which means a large share of the "cold" effect is really the wind that arrives with it. The remaining gap points the same way but no longer clears significance on its own.
If passing falls, why doesn't scoring fall?
That is exactly what our temperature and scoring study found: team scoring is essentially flat from freezing to 85°F. The two results fit together — cold suppresses passing efficiency, and teams compensate with volume, field position and the running game rather than simply scoring less. Cold changes how offences move the ball more than how many points they finish with.
Do interceptions rise in the cold?
No. Interceptions and sacks sit flat against baseline in every band. Cold makes passing less efficient; it does not make quarterbacks careless — the same pattern we measured in wind.
Can I bet passing-yard props off this?
Not directly. This study measures performance against a quarterback's own form, not against the posted prop line. We hold no historical archive of prop lines, so we cannot say whether the market already prices cold in. Until we do, this is context for reading a number, not a signal to bet it.